Customizing THREDDS Data Server Web Interfaces to Promote Data Discovery and Accessibility

The recent paradigm shift toward open data has encouraged the growth of freely-available, web-based data repositories to the point of overwhelming the typical user. Accordingly, the responsibilities of open-data stewardship have expanded to ensuring data is not only available, but also accessible to those who would benefit from its use. This study introduces new capabilities in the customization and management of THREDDS Data Server (TDS) instances as a method of improving data accessibility. The TDS is a web server used to provide access to a broad spectrum of scientific datasets via catalogs of metadata; because the data served by instances of the TDS can vary widely, fine-tuned control over user interfaces facilitates discoverability and ease of access. Recent enhancements in TDS administrative control include customizable page design via user-contributed CSS, as well as use of an HTML templating engine to enable the display of extensible metadata relevant to various datasets and catalogs. TDS administrators may also choose to provide Jupyter Notebook files, which are registered as data viewers for a subset of datasets on the server and which can be downloaded from the web service by the end user. Jupyter Notebook viewers are a mixture of Markdown and Python scripts which contain information about the data or data access, processing, and analysis methods. Although Notebook viewers are primarily intended to promote simple data visualization without requiring massive data transfers, they may also be targeted towards educational exercises or verifying and reproducing techniques presented in publications. These new features in the TDS web interface improve the navigability of large data repositories and the discoverability of the data they contain, lowering the barrier-to-entry on data access. Although open data is inherently available to everyone, it can be made accessible to a broader audience by enabling data providers to implement dataset or usage-specific customization on top of a core infrastructure of data interfaces.